Frosting　is　an　undesired　phenomenon　for　air　source　heat　pumps　(ASHPs),　and　numerical　correlations　are　valuable　for　predicting　the　dynamic　frosting　properties.　However,　the　applicable　ranges　of　current　correlations　greatly　deviate　from　the　practical　frosting　conditions　of　ASHPs.　To　solve　the　above-mentioned　problem,　this　paper　develops　a　set　of　semiempirical　dimensionless　correlations　under　the　practical　frosting　conditions　of　ASHPs.　The　sample　data　is　generated　from　an　improved　generalized　frost　formation　model.　With　these　data,　semiempirical　dimensionless　correlations　are　developed　by　the　multiple　linear　regression　method.　Frosting　properties　are　regressed　as　functions　of　the　Reynolds　number,　Fourier　number,　relative　humidity,　and　of　dimensionless　temperature.　The　proposed　correlations　are　validated　by　the　sample　data　within　an　error　of　+/-　10%　and　examined　by　experimental　data　from　open　literature　with　good　agreements　of　+/-　15%.　The　results　indicate　that　the　proposed　semiempirical　dimensionless　correlations　can　predict　the　frosting　properties　under　practical　frosting　conditions　of　ASHPs.